{"metadata":{"kernelspec":{"language":"python","display_name":"Python 3","name":"python3"},"language_info":{"name":"python","version":"3.10.13","mimetype":"text/x-python","codemirror_mode":{"name":"ipython","version":3},"pygments_lexer":"ipython3","nbconvert_exporter":"python","file_extension":".py"},"kaggle":{"accelerator":"none","dataSources":[{"sourceId":70203,"databundleVersionId":8068726,"sourceType":"competition"}],"dockerImageVersionId":30698,"isInternetEnabled":false,"language":"python","sourceType":"notebook","isGpuEnabled":false}},"nbformat_minor":4,"nbformat":4,"cells":[{"cell_type":"code","source":"import os\nos.environ[\"KERAS_BACKEND\"] = \"jax\"  # \"jax\" or \"tensorflow\" or \"torch\" \n\nimport keras_cv\nimport keras\nimport keras.backend as K\nimport tensorflow as tf\nimport tensorflow_io as tfio\n\nimport numpy as np \nimport pandas as pd\n\nfrom glob import glob\nfrom tqdm import tqdm\n\nimport librosa\nimport IPython.display as ipd\nimport librosa.display as lid\n\nimport matplotlib.pyplot as plt\nimport matplotlib as mpl\n\ncmap = mpl.cm.get_cmap('coolwarm')","metadata":{"_uuid":"8f2839f25d086af736a60e9eeb907d3b93b6e0e5","_cell_guid":"b1076dfc-b9ad-4769-8c92-a6c4dae69d19","execution":{"iopub.status.busy":"2024-06-08T10:22:23.114137Z","iopub.execute_input":"2024-06-08T10:22:23.114454Z","iopub.status.idle":"2024-06-08T10:22:40.106297Z","shell.execute_reply.started":"2024-06-08T10:22:23.114424Z","shell.execute_reply":"2024-06-08T10:22:40.104979Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"print(\"TensorFlow:\", tf.__version__)\nprint(\"Keras:\", keras.__version__)\nprint(\"KerasCV:\", keras_cv.__version__)","metadata":{"execution":{"iopub.status.busy":"2024-06-08T10:22:40.108216Z","iopub.execute_input":"2024-06-08T10:22:40.109256Z","iopub.status.idle":"2024-06-08T10:22:40.114137Z","shell.execute_reply.started":"2024-06-08T10:22:40.109227Z","shell.execute_reply":"2024-06-08T10:22:40.113126Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"class CFG:\n    seed = 42\n    \n    # Input image size and batch size\n    img_size = [128, 384]\n    batch_size = 64\n    \n    # Audio duration, sample rate, and length\n    duration = 15 # second\n    sample_rate = 32000\n    audio_len = duration*sample_rate\n    \n    # STFT parameters\n    nfft = 2028\n    window = 2048\n    hop_length = audio_len // (img_size[1] - 1)\n    fmin = 20\n    fmax = 16000\n    \n    # Number of epochs, model name\n    epochs = 10\n       \n    # Data augmentation parameters\n    augment=True\n\n    # Class Labels for BirdCLEF 24\n    class_names = sorted(os.listdir('/kaggle/input/birdclef-2024/train_audio/'))\n    num_classes = len(class_names)\n    class_labels = list(range(num_classes))\n    label2name = dict(zip(class_labels, class_names))\n    name2label = {v:k for k,v in label2name.items()}","metadata":{"execution":{"iopub.status.busy":"2024-06-08T10:22:40.115301Z","iopub.execute_input":"2024-06-08T10:22:40.115689Z","iopub.status.idle":"2024-06-08T10:22:40.250060Z","shell.execute_reply.started":"2024-06-08T10:22:40.115656Z","shell.execute_reply":"2024-06-08T10:22:40.249351Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"BASE_PATH = '/kaggle/input/birdclef-2024'\n","metadata":{"execution":{"iopub.status.busy":"2024-06-08T10:22:42.214613Z","iopub.execute_input":"2024-06-08T10:22:42.214971Z","iopub.status.idle":"2024-06-08T10:22:42.219308Z","shell.execute_reply.started":"2024-06-08T10:22:42.214942Z","shell.execute_reply":"2024-06-08T10:22:42.218268Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"df = pd.read_csv(f'{BASE_PATH}/train_metadata.csv')\ndf['filepath'] = BASE_PATH + '/train_audio/' + df.filename\ndf['target'] = df.primary_label.map(CFG.name2label)\ndf['filename'] = df.filepath.map(lambda x: x.split('/')[-1])\ndf['xc_id'] = df.filepath.map(lambda x: x.split('/')[-1].split('.')[0])\n\n# Display rwos\ndf.head(2)\n","metadata":{"execution":{"iopub.status.busy":"2024-06-08T10:22:43.157178Z","iopub.execute_input":"2024-06-08T10:22:43.157553Z","iopub.status.idle":"2024-06-08T10:22:43.401350Z","shell.execute_reply.started":"2024-06-08T10:22:43.157522Z","shell.execute_reply":"2024-06-08T10:22:43.400242Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def load_audio(filepath):\n    audio, sr = librosa.load(filepath)\n    return audio, sr\n\ndef get_spectrogram(audio):\n    spec = librosa.feature.melspectrogram(y=audio, \n                                   sr=CFG.sample_rate, \n                                   n_mels=256,\n                                   n_fft=2048,\n                                   hop_length=512,\n                                   fmax=CFG.fmax,\n                                   fmin=CFG.fmin,\n                                   )\n    spec = librosa.power_to_db(spec, ref=1.0)\n    min_ = spec.min()\n    max_ = spec.max()\n    if max_ != min_:\n        spec = (spec - min_)/(max_ - min_)\n    return spec\n\ndef display_audio(row):\n    # Caption for viz\n    caption = f'Id: {row.filename} | Name: {row.common_name} | Sci.Name: {row.scientific_name} | Rating: {row.rating}'\n    # Read audio file\n    audio, sr = load_audio(row.filepath)\n    # Keep fixed length audio\n    audio = audio[:CFG.audio_len]\n    # Spectrogram from audio\n    spec = get_spectrogram(audio)\n    # Display audio\n    print(\"# Audio:\")\n    display(ipd.Audio(audio, rate=CFG.sample_rate))\n    print('# Visualization:')\n    fig, ax = plt.subplots(2, 1, figsize=(12, 2*3), sharex=True, tight_layout=True)\n    fig.suptitle(caption)\n    # Waveplot\n    lid.waveshow(audio,\n                 sr=CFG.sample_rate,\n                 ax=ax[0],\n                 color= cmap(0.1))\n    # Specplot\n    lid.specshow(spec, \n                 sr = CFG.sample_rate, \n                 hop_length=512,\n                 n_fft=2048,\n                 fmin=CFG.fmin,\n                 fmax=CFG.fmax,\n                 x_axis = 'time', \n                 y_axis = 'mel',\n                 cmap = 'coolwarm',\n                 ax=ax[1])\n    ax[0].set_xlabel('');\n    fig.show()","metadata":{"execution":{"iopub.status.busy":"2024-06-08T10:22:44.314469Z","iopub.execute_input":"2024-06-08T10:22:44.314880Z","iopub.status.idle":"2024-06-08T10:22:44.326887Z","shell.execute_reply.started":"2024-06-08T10:22:44.314847Z","shell.execute_reply":"2024-06-08T10:22:44.326033Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"row = df.iloc[35]\n\n# Display audio\ndisplay_audio(row)","metadata":{"execution":{"iopub.status.busy":"2024-06-08T10:22:46.140491Z","iopub.execute_input":"2024-06-08T10:22:46.140818Z","iopub.status.idle":"2024-06-08T10:22:58.303725Z","shell.execute_reply.started":"2024-06-08T10:22:46.140793Z","shell.execute_reply":"2024-06-08T10:22:58.302662Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Import required packages\nfrom sklearn.model_selection import train_test_split\n\ntrain_df, valid_df = train_test_split(df, test_size=0.2)\n\nprint(f\"Num Train: {len(train_df)} | Num Valid: {len(valid_df)}\")","metadata":{"execution":{"iopub.status.busy":"2024-06-08T10:22:58.305442Z","iopub.execute_input":"2024-06-08T10:22:58.305992Z","iopub.status.idle":"2024-06-08T10:22:58.469402Z","shell.execute_reply.started":"2024-06-08T10:22:58.305966Z","shell.execute_reply":"2024-06-08T10:22:58.468453Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Decodes Audio\ndef build_decoder(with_labels=True, dim=1024):\n    def get_audio(filepath):\n        file_bytes = tf.io.read_file(filepath)\n        audio = tfio.audio.decode_vorbis(file_bytes)  # decode .ogg file\n        audio = tf.cast(audio, tf.float32)\n        if tf.shape(audio)[1] > 1:  # stereo -> mono\n            audio = audio[..., 0:1]\n        audio = tf.squeeze(audio, axis=-1)\n        return audio\n\n    def crop_or_pad(audio, target_len, pad_mode=\"constant\"):\n        audio_len = tf.shape(audio)[0]\n        diff_len = abs(\n            target_len - audio_len\n        )  # find difference between target and audio length\n        if audio_len < target_len:  # do padding if audio length is shorter\n            pad1 = tf.random.uniform([], maxval=diff_len, dtype=tf.int32)\n            pad2 = diff_len - pad1\n            audio = tf.pad(audio, paddings=[[pad1, pad2]], mode=pad_mode)\n        elif audio_len > target_len:  # do cropping if audio length is larger\n            idx = tf.random.uniform([], maxval=diff_len, dtype=tf.int32)\n            audio = audio[idx : (idx + target_len)]\n        return tf.reshape(audio, [target_len])\n\n    def apply_preproc(spec):\n        # Standardize\n        mean = tf.math.reduce_mean(spec)\n        std = tf.math.reduce_std(spec)\n        spec = tf.where(tf.math.equal(std, 0), spec - mean, (spec - mean) / std)\n\n        # Normalize using Min-Max\n        min_val = tf.math.reduce_min(spec)\n        max_val = tf.math.reduce_max(spec)\n        spec = tf.where(\n            tf.math.equal(max_val - min_val, 0),\n            spec - min_val,\n            (spec - min_val) / (max_val - min_val),\n        )\n        return spec\n\n    def get_target(target):\n        target = tf.reshape(target, [1])\n        target = tf.cast(tf.one_hot(target, CFG.num_classes), tf.float32)\n        target = tf.reshape(target, [CFG.num_classes])\n        return target\n\n    def decode(path):\n        # Load audio file\n        audio = get_audio(path)\n        # Crop or pad audio to keep a fixed length\n        audio = crop_or_pad(audio, dim)\n        # Audio to Spectrogram\n        spec = keras.layers.MelSpectrogram(\n            num_mel_bins=CFG.img_size[0],\n            fft_length=CFG.nfft,\n            sequence_stride=CFG.hop_length,\n            sampling_rate=CFG.sample_rate,\n        )(audio)\n        # Apply normalization and standardization\n        spec = apply_preproc(spec)\n        # Spectrogram to 3 channel image (for imagenet)\n        spec = tf.tile(spec[..., None], [1, 1, 3])\n        spec = tf.reshape(spec, [*CFG.img_size, 3])\n        return spec\n\n    def decode_with_labels(path, label):\n        label = get_target(label)\n        return decode(path), label\n\n    return decode_with_labels if with_labels else decode","metadata":{"execution":{"iopub.status.busy":"2024-06-08T10:22:58.470551Z","iopub.execute_input":"2024-06-08T10:22:58.470813Z","iopub.status.idle":"2024-06-08T10:22:58.487019Z","shell.execute_reply.started":"2024-06-08T10:22:58.470790Z","shell.execute_reply":"2024-06-08T10:22:58.486128Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def build_augmenter():\n    augmenters = [\n        keras_cv.layers.MixUp(alpha=0.4),\n        keras_cv.layers.RandomCutout(height_factor=(1.0, 1.0),\n                                     width_factor=(0.06, 0.12)), # time-masking\n        keras_cv.layers.RandomCutout(height_factor=(0.06, 0.1),\n                                     width_factor=(1.0, 1.0)), # freq-masking\n    ]\n    \n    def augment(img, label):\n        data = {\"images\":img, \"labels\":label}\n        for augmenter in augmenters:\n            if tf.random.uniform([]) < 0.35:\n                data = augmenter(data, training=True)\n        return data[\"images\"], data[\"labels\"]\n    \n    return augment","metadata":{"execution":{"iopub.status.busy":"2024-06-08T10:22:58.489513Z","iopub.execute_input":"2024-06-08T10:22:58.489855Z","iopub.status.idle":"2024-06-08T10:22:58.498276Z","shell.execute_reply.started":"2024-06-08T10:22:58.489823Z","shell.execute_reply":"2024-06-08T10:22:58.497393Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def build_dataset(paths, labels=None, batch_size=32, \n                  decode_fn=None, augment_fn=None, cache=True,\n                  augment=False, shuffle=2048):\n\n    if decode_fn is None:\n        decode_fn = build_decoder(labels is not None, dim=CFG.audio_len)\n\n    if augment_fn is None:\n        augment_fn = build_augmenter()\n        \n    AUTO = tf.data.experimental.AUTOTUNE\n    slices = (paths,) if labels is None else (paths, labels)\n    ds = tf.data.Dataset.from_tensor_slices(slices)\n    ds = ds.map(decode_fn, num_parallel_calls=AUTO)\n    ds = ds.cache() if cache else ds\n    if shuffle:\n        opt = tf.data.Options()\n        ds = ds.shuffle(shuffle, seed=CFG.seed)\n        opt.experimental_deterministic = False\n        ds = ds.with_options(opt)\n    ds = ds.batch(batch_size, drop_remainder=True)\n    ds = ds.map(augment_fn, num_parallel_calls=AUTO) if augment else ds\n    ds = ds.prefetch(AUTO)\n    return ds","metadata":{"execution":{"iopub.status.busy":"2024-06-08T10:22:58.499368Z","iopub.execute_input":"2024-06-08T10:22:58.500101Z","iopub.status.idle":"2024-06-08T10:22:58.508216Z","shell.execute_reply.started":"2024-06-08T10:22:58.500076Z","shell.execute_reply":"2024-06-08T10:22:58.507359Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Train\ntrain_paths = train_df.filepath.values\ntrain_labels = train_df.target.values\ntrain_ds = build_dataset(train_paths, train_labels, batch_size=CFG.batch_size,\n                         shuffle=True, augment=CFG.augment)\n\n# Valid\nvalid_paths = valid_df.filepath.values\nvalid_labels = valid_df.target.values\nvalid_ds = build_dataset(valid_paths, valid_labels, batch_size=CFG.batch_size,\n                         shuffle=False, augment=False)","metadata":{"execution":{"iopub.status.busy":"2024-06-08T10:22:58.509227Z","iopub.execute_input":"2024-06-08T10:22:58.510046Z","iopub.status.idle":"2024-06-08T10:23:02.630848Z","shell.execute_reply.started":"2024-06-08T10:22:58.510021Z","shell.execute_reply":"2024-06-08T10:23:02.630069Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def plot_batch(batch, row=3, col=3, label2name=None,):\n    \"\"\"Plot one batch data\"\"\"\n    if isinstance(batch, tuple) or isinstance(batch, list):\n        specs, tars = batch\n    else:\n        specs = batch\n        tars = None\n    plt.figure(figsize=(col*5, row*3))\n    for idx in range(row*col):\n        ax = plt.subplot(row, col, idx+1)\n        lid.specshow(np.array(specs[idx, ..., 0]), \n                     n_fft=CFG.nfft, \n                     hop_length=CFG.hop_length, \n                     sr=CFG.sample_rate,\n                     x_axis='time',\n                     y_axis='mel',\n                     cmap='coolwarm')\n        if tars is not None:\n            label = tars[idx].numpy().argmax()\n            name = label2name[label]\n            plt.title(name)\n    plt.tight_layout()\n    plt.show()","metadata":{"execution":{"iopub.status.busy":"2024-06-08T10:23:02.631850Z","iopub.execute_input":"2024-06-08T10:23:02.632099Z","iopub.status.idle":"2024-06-08T10:23:02.639992Z","shell.execute_reply.started":"2024-06-08T10:23:02.632076Z","shell.execute_reply":"2024-06-08T10:23:02.639118Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"from keras import Model\nfrom keras.callbacks import LearningRateScheduler\nfrom keras import layers","metadata":{"execution":{"iopub.status.busy":"2024-06-08T10:23:02.641075Z","iopub.execute_input":"2024-06-08T10:23:02.641332Z","iopub.status.idle":"2024-06-08T10:23:02.651786Z","shell.execute_reply.started":"2024-06-08T10:23:02.641310Z","shell.execute_reply":"2024-06-08T10:23:02.650944Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"def naive_inception_module(layer_in, f1, f2, f3):\n    # 5x9 conv\n    conv1_1 = layers.Conv2D(f1, (1,1), padding='same', activation=layers.LeakyReLU(negative_slope=0.25),strides=4,kernel_initializer='random_normal',bias_initializer='zeros')(layer_in)\n    # 7x7 conv\n    conv1_2 = layers.Conv2D(f2, (3,3), padding='same', activation=layers.LeakyReLU(negative_slope=0.25),strides=4,kernel_initializer='random_normal',bias_initializer='zeros')(layer_in)\n    # 9x5 conv\n    conv1_3 = layers.Conv2D(f3, (5,5), padding='same', activation=layers.LeakyReLU(negative_slope=0.25),strides=4,kernel_initializer='random_normal',bias_initializer='zeros')(layer_in)\n    \n    # concatenate filters, assumes filters/channels last\n    layer_out = layers.concatenate([conv1_1, conv1_2, conv1_3], axis=-1)\n    return layer_out\n \n# define model2 input\nvisible = layers.Input(shape=(128, 384, 3))\n# add inception module\nlayer = naive_inception_module(visible, 4, 4, 4)\n\nconv2=layers.Conv2D(filters=16, kernel_size=(1,1),padding='valid', activation=layers.LeakyReLU(negative_slope=0.25),strides=1,kernel_initializer='random_normal',bias_initializer='zeros')(layers.BatchNormalization()(layer))\nmaxpool2=layers.MaxPooling2D(pool_size=(2, 2),strides=1)(conv2)\nnorm2=(layers.BatchNormalization()(maxpool2))\n\nconv3=layers.Conv2D(filters=32, kernel_size=(3,3),padding='valid', \n                    activation=layers.LeakyReLU(negative_slope=0.25),strides=2,kernel_initializer='random_normal',bias_initializer='ones')(norm2)\nmaxpool3=layers.MaxPooling2D(pool_size=(2, 2),strides=1)(conv3)\nnorm3=(layers.BatchNormalization()(maxpool3))\n\nconv4=layers.Conv2D(filters=64, kernel_size=(3,3),padding='valid', \n                    activation=layers.LeakyReLU(negative_slope=0.25),strides=2,kernel_initializer='random_normal',bias_initializer='ones')(norm3)\nmaxpool4=layers.MaxPooling2D(pool_size=(2, 2),strides=1)(conv4)\nnorm4=(layers.BatchNormalization()(maxpool4))\n\nconv5=layers.Conv2D(filters=64, kernel_size=(5,5),padding='valid', \n                    activation=layers.LeakyReLU(negative_slope=0.25),strides=2,kernel_initializer='random_normal',bias_initializer='ones')(norm4)\nmaxpool5=layers.MaxPooling2D(pool_size=(2, 2),strides=1)(conv5)\nnorm5=(layers.BatchNormalization()(conv5))\n\nflat=layers.Flatten()(norm5)\n\nFC1=layers.Dense(64, activation=layers.LeakyReLU(negative_slope=0.25),kernel_initializer='random_normal',bias_initializer='ones')(flat)\ndrop_FC1=layers.Dropout((0.5))(layers.BatchNormalization()(FC1))\nFC2=layers.Dense(32, activation=layers.LeakyReLU(negative_slope=0.25),kernel_initializer='random_normal',bias_initializer='ones')(drop_FC1)\ndrop_FC2=drop3=layers.Dropout((0.5))(layers.BatchNormalization()(FC2))\n\nFC3=layers.Dense(16, activation=layers.LeakyReLU(negative_slope=0.25),kernel_initializer='random_normal',bias_initializer='ones')(drop_FC2)\ndrop_FC3=drop3=layers.Dropout((0.5))(layers.BatchNormalization()(FC3))\n\n\nlayer_out=layers.Dense(CFG.num_classes, activation='softmax',kernel_initializer='random_normal',bias_initializer='zeros')(drop_FC3)\n\n# create model2\nmodel2 = Model(inputs=visible, outputs=layer_out)\n\n# summarize model2\nmodel2.summary()\n\n\n\n","metadata":{"execution":{"iopub.status.busy":"2024-06-08T10:23:02.652942Z","iopub.execute_input":"2024-06-08T10:23:02.653198Z","iopub.status.idle":"2024-06-08T10:23:07.761832Z","shell.execute_reply.started":"2024-06-08T10:23:02.653175Z","shell.execute_reply":"2024-06-08T10:23:07.760955Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"model2.compile(optimizer=\"adam\",\n              loss=keras.losses.CategoricalCrossentropy(label_smoothing=0.02),\n              metrics=[keras.metrics.AUC(name='auc')],)","metadata":{"execution":{"iopub.status.busy":"2024-06-08T10:23:09.252343Z","iopub.execute_input":"2024-06-08T10:23:09.252769Z","iopub.status.idle":"2024-06-08T10:23:09.397523Z","shell.execute_reply.started":"2024-06-08T10:23:09.252736Z","shell.execute_reply":"2024-06-08T10:23:09.396535Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"import math\nimport matplotlib.pyplot as plt\nimport numpy as np\nimport keras\n\ndef get_lr_callback(batch_size=8, mode='step', epochs=150, plot=False):\n    # Define learning rate parameters\n    lr_start = 1e-2   \n    lr_max = lr_start*0.9   \n    lr_min = 1e-5\n    lr_ramp_ep, lr_sus_ep, lr_decay = 3, 0, 0.9  # Changed decay to 0.1 for significant reduction\n    step_size = 20  # Reduce learning rate every 10 epochs\n\n    def lrfn(epoch):  # Learning rate update function\n        if epoch < lr_ramp_ep:\n            lr = (lr_max - lr_start) / lr_ramp_ep * epoch + lr_start\n        elif epoch < lr_ramp_ep + lr_sus_ep:\n            lr = lr_max\n        elif mode == 'exp':\n            lr = (lr_max - lr_min) * lr_decay**(epoch - lr_ramp_ep - lr_sus_ep) + lr_min\n        elif mode == 'step':\n            lr = lr_max * lr_decay**((epoch - lr_ramp_ep - lr_sus_ep) // step_size)\n        elif mode == 'cos':\n            decay_total_epochs = epochs - lr_ramp_ep - lr_sus_ep + 3\n            decay_epoch_index = epoch - lr_ramp_ep - lr_sus_ep\n            phase = math.pi * decay_epoch_index / decay_total_epochs\n            lr = (lr_max - lr_min) * 0.5 * (1 + math.cos(phase)) + lr_min\n        return lr\n\n    if plot:  # Plot lr curve if plot is True\n        plt.figure(figsize=(10, 5))\n        plt.plot(np.arange(epochs), [lrfn(epoch) for epoch in np.arange(epochs)], marker='o')\n        plt.xlabel('epoch')\n        plt.ylabel('lr')\n        plt.title('LR Scheduler')\n        plt.show()\n\n    return keras.callbacks.LearningRateScheduler(lrfn, verbose=False)  # Create lr callback\n","metadata":{"execution":{"iopub.status.busy":"2024-06-08T10:23:13.401769Z","iopub.execute_input":"2024-06-08T10:23:13.402632Z","iopub.status.idle":"2024-06-08T10:23:13.413718Z","shell.execute_reply.started":"2024-06-08T10:23:13.402598Z","shell.execute_reply":"2024-06-08T10:23:13.412796Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"lr_cb = get_lr_callback(CFG.batch_size, plot=True)","metadata":{"execution":{"iopub.status.busy":"2024-06-08T10:23:14.112166Z","iopub.execute_input":"2024-06-08T10:23:14.112542Z","iopub.status.idle":"2024-06-08T10:23:14.371112Z","shell.execute_reply.started":"2024-06-08T10:23:14.112513Z","shell.execute_reply":"2024-06-08T10:23:14.370237Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"ckpt_cb = keras.callbacks.ModelCheckpoint(\"AKCNN_model1.weights.h5\",\n                                         monitor='val_auc',\n                                         save_best_only=True,\n                                         save_weights_only=True,\n                                         mode='max')","metadata":{"execution":{"iopub.status.busy":"2024-06-08T10:23:15.561754Z","iopub.execute_input":"2024-06-08T10:23:15.562113Z","iopub.status.idle":"2024-06-08T10:23:15.567052Z","shell.execute_reply.started":"2024-06-08T10:23:15.562084Z","shell.execute_reply":"2024-06-08T10:23:15.566081Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"# Train the model\nhistory = model2.fit(\n    train_ds, \n    validation_data=valid_ds, \n    epochs=50,\n    callbacks=[lr_cb, ckpt_cb], \n    verbose=1\n)\n\n","metadata":{"execution":{"iopub.status.busy":"2024-06-08T10:23:20.138161Z","iopub.execute_input":"2024-06-08T10:23:20.138558Z","iopub.status.idle":"2024-06-08T11:27:18.954633Z","shell.execute_reply.started":"2024-06-08T10:23:20.138526Z","shell.execute_reply":"2024-06-08T11:27:18.953688Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{"execution":{"iopub.status.busy":"2024-06-08T11:32:19.144812Z","iopub.execute_input":"2024-06-08T11:32:19.145236Z","iopub.status.idle":"2024-06-08T11:32:19.151354Z","shell.execute_reply.started":"2024-06-08T11:32:19.145200Z","shell.execute_reply":"2024-06-08T11:32:19.150330Z"},"trusted":true},"execution_count":null,"outputs":[]},{"cell_type":"code","source":"","metadata":{},"execution_count":null,"outputs":[]}]}